Assignment 2

library(data.table)
library(leaflet)
library(RColorBrewer)
library(stringr)
library(geosphere)
library(geojsonR)
library(rgdal)
library(ggmap)
library(dplyr)
library(tidyr)
library(rgeos)
library(shiny)

setwd('/Users/greatyifan/Desktop/@Columbia/2020spring/2_DataViz/course_materials/Exercises/07_fire')

#read in data
df_fire <- fread('building_fires.csv')
df_house <- fread('FDNY_Firehouse_Listing.csv')

1. Location of Severe Fires

# levels(df_fire$HIGHEST_LEVEL_DESC)
# it appears that the levels replicate themselves with with minor description differences, so combine it first.
df_fire[, HIGHEST_LEVEL_DESC := ifelse(HIGHEST_LEVEL_DESC == "11 - First Alarm", "1 - More than initial alarm, less than Signal 7-5" , HIGHEST_LEVEL_DESC)]
df_fire[, HIGHEST_LEVEL_DESC := ifelse(HIGHEST_LEVEL_DESC == "22 - Second Alarm", "2 - 2nd alarm" , HIGHEST_LEVEL_DESC)]
df_fire[, HIGHEST_LEVEL_DESC := ifelse(HIGHEST_LEVEL_DESC == "33 - Third Alarm", "3 - 3rd alarm" , HIGHEST_LEVEL_DESC)]
df_fire[, HIGHEST_LEVEL_DESC := ifelse(HIGHEST_LEVEL_DESC == "44 - Fourth Alarm", "4 - 4th alarm" , HIGHEST_LEVEL_DESC)]
df_fire[, HIGHEST_LEVEL_DESC := ifelse(HIGHEST_LEVEL_DESC == "55 - Fifth Alarm", "5 - 5th alarm" , HIGHEST_LEVEL_DESC)]
df_fire[, HIGHEST_LEVEL_DESC := ifelse(HIGHEST_LEVEL_DESC == "75 - All Hands Working", "7 - Signal 7-5" , HIGHEST_LEVEL_DESC)]

df_fire$HIGHEST_LEVEL_DESC <- factor(df_fire$HIGHEST_LEVEL_DESC)

# cast date-time column into data-time type data
df_fire$ARRIVAL_DATE_TIME <- as.POSIXct(df_fire$ARRIVAL_DATE_TIME, 
                                        format = '%m/%d/%Y %I:%M:%S %p') 
df_fire$INCIDENT_DATE_TIME <- as.POSIXct(df_fire$INCIDENT_DATE_TIME, 
                                         format = '%m/%d/%Y %I:%M:%S %p')

# attribution to mapbox
attr <- "© <a href='https://github.com/ChengweiWang3210'>Chengwei Wang</a>"

# set up a base map
base_map <- leaflet(options = leafletOptions(minZoom = 10, maxZoom = 18)) %>% 
#fix the zoom level so that zoom out of new york too far is not optional.
  addTiles(attribution = attr) %>%
  setView(zoom = 10, lng = -74.00919, lat = 40.69999) %>%
  addProviderTiles(provider = "CartoDB.VoyagerNoLabels")
  
df_highest <- subset(df_fire, df_fire$HIGHEST_LEVEL_DESC == '7 - Signal 7-5')

# add on incident points and popups
base_map %>%
  addCircles(data = df_highest,
             lng = ~lon, lat = ~lat, radius = .1,
             stroke = .5, color = 'red', fillOpacity = .01, 
             popup = paste0('Address: ', df_highest$address, '<br/>', 
                            'Incident Data: ', df_highest$INCIDENT_DATE_TIME, '<br/>',
                            'Total Incident Duration: ',
                            round(df_highest$TOTAL_INCIDENT_DURATION/60),
                            ' minutes'))

NA

2. Layers and Clusters

a) Color by Type of Property

# recategorize property types 

## unique(df_highest$PROPERTY_USE_DESC)[substr(unique(df_highest$PROPERTY_USE_DESC), 1,3 ) < 200]
## unique(df_highest$PROPERTY_USE_DESC)[substr(unique(df_highest$PROPERTY_USE_DESC), 1,3 ) > 200]
list_property <- unique(df_highest$PROPERTY_USE_DESC)
# order of the following codes matters.
list_property[str_detect(tolower(list_property),'store|shop|club|business|cafe|retail|warehouse|sales|service')] <- 'Business Sphere'
list_property[str_detect(tolower(list_property), 'doctor|clinic|hospital|recovery|nursing|care|sanita')] <- 'Medical'
list_property[str_detect(tolower(list_property),'playground|open| 
|street|terminal|lot|bus|pier|outside|yard|processing|recreation|drinking|parking|shed|construction|distribution|aircraft')] <- 'Open Aera'
list_property[str_detect(tolower(list_property), 'family|residential,')] <- 'Residence'
list_property[str_detect(tolower(list_property), 'hotel|dorm|cleaning|storage|shelter|property')] <- 'Dorms, Shelters, Hotels'
list_property[str_detect(tolower(list_property), 'educ|school')] <- 'Schools'
list_property[str_detect(tolower(list_property), 'church|hospices|station|arena|assembly|theater|museum|parlor|office|public|bank|hall|disability|studio|center|court|plant|gym|lab|cleaning|storage|property')] <- 'Public 
'
list_property[str_detect(tolower(list_property), 'undetermined|none')] <- 'Undefined'

# combine recoded property categories with original property_use_desc columns
df_combine <- cbind(unique(df_highest$PROPERTY_USE_DESC), list_property)
colnames(df_combine) <- c('PROPERTY_USE_DESC', 'property')
df_combine <- as.data.frame(df_combine)

# join the recoded property column back to the dataframe
df_fire_property <- left_join(df_fire, df_combine, by = 'PROPERTY_USE_DESC')

# rank the "property" variable's level by the number of incidents falling into these categories
ranked <- sort(table(df_fire_property$property), decreasing = T)
df_fire_property$property <- factor(df_fire_property$property, levels = names(ranked))
## above 2 lines of code are trying to ranking types of property by their frequency, and use this to show a more imformative legend in the following map. 

# brew the colors for the property variable
colors <- brewer.pal(uniqueN(df_fire_property$property), "Set2")
propCol <- colorFactor(colors, df_fire_property$property)

# pick out data with highest level of alarm
df_highest_property <- subset(df_fire_property, 
                              df_fire_property$HIGHEST_LEVEL_DESC == '7 - Signal 7-5')


base_map %>%
  addCircles(data = df_highest_property,
             lng = ~lon, lat = ~lat, radius = 1, color = ~propCol(property),
             weight = 1, stroke = 1, fillOpacity = .7, 
             popup = paste0('Address: ', df_highest_property$address, '<br/>', 
                            'Incident Data: ', df_highest_property$INCIDENT_DATE_TIME, '<br/>',
                            'Total Incident Duration: ',
                            round(df_highest_property$TOTAL_INCIDENT_DURATION/60),
                            ' minutes<br/>',
                            'PropertyType: ', df_highest_property$property)) %>%
  addLegend(data = df_highest_property, group = 'Incidents',
            title = "Property Types", position = "topleft",
            pal = propCol, values = ~property)

b) Cluster

base_map %>%
  addCircleMarkers(data = df_highest_property,
             lng = ~lon, lat = ~lat, radius = .1, color = ~propCol(property),
             stroke = 0, fillOpacity = .9, 
             popup = paste0('Address: ', df_highest_property$address, '<br/>', 
                            'Incident Data: ', df_highest_property$INCIDENT_DATE_TIME, '<br/>',
                            'Total Incident Duration: ',
                            round(df_highest_property$TOTAL_INCIDENT_DURATION/60),
                            ' minutes<br/>',
                            'PropertyType: ', df_highest_property$property), 
             clusterOptions = markerClusterOptions(spiderfyOnMaxZoom = 10)) %>%
  addLegend(data = df_highest_property, group = 'Incidents',
            title = "Property Types", position = "topleft",
            pal = propCol, values = ~property)

3. Fire Houses

# add on an icon png
house_icon <- icons(iconUrl = 
  "/Users/greatyifan/Desktop/@Columbia/2020spring/2_DataViz/assignment/house-icon.png",
                    iconWidth = 8, iconHeight = 8)

base_map %>%
  addCircleMarkers(data = df_highest_property, group = 'Incidents', 
                   lng = ~lon, lat = ~lat, 
                   radius = df_highest_property$UNITS_ONSCENE/4,
                   color = ~propCol(property),
                   stroke = 0, weight = 0, fillOpacity = .7, 
                   popup = ~paste0('Address: ', df_highest_property$address, '<br/>', 
                            'Incident Data: ', df_highest_property$INCIDENT_DATE_TIME, '<br/>',
                            'Total Incident Duration: ',
                            round(df_highest_property$TOTAL_INCIDENT_DURATION/60), 
                            ' minutes<br/>',
                            'PropertyType: ', df_highest_property$property)) %>%
  addLegend(data = df_highest_property, group = 'Incidents',
            pal = propCol, values = ~property, 
            title = 'Property Types', position = 'topleft') %>%
  addMarkers(data = df_house, group = 'Firehouses', 
             lng = ~Longitude, lat = ~Latitude,
             icon = house_icon, 
             popup = ~paste0('Address: ', df_house$FacilityAddress, '<br/>', 
                            'Borough: ', df_house$Borough)) %>%
  addLayersControl(baseGroups = 'openStreetNYC',
                   overlayGroups = c('Incidents','Firehouses'), 
                   options = layersControlOptions(collapsed = T))

4. Distance from Firehouse and Response Time

a) Calculate Distance

# this function returns nrow(x) * nrow(y) matrix, where the i-th column indicates the distances between each geopoint in x with the i-th point in y. Similarly, each element in the j-th row means the distance between the j-th point in x with each point in y. In our case, if we want to find out the nearest firehouse for a certain point, we have to find the minimum element for each row, and return the number of column where the minimum point is in, which is the nearest firehouse for that incident. 

mx_dist <- distm(x = matrix(data = c(df_fire$lon, df_fire$lat), ncol = 2),
      y = matrix(data = c(df_house$Longitude, df_house$Latitude), ncol = 2), 
      fun = distGeo)

min_dist <- apply(mx_dist, 1, min, na.rm = T)

nearest_house <- apply(mx_dist, 1, function(x)which(x == min(x, na.rm = T)))

# nrow(df_house) # we have 218 fire houses

# summary(nearest_house) # everything seems right

df_fire_property$min_dist <- min_dist
df_fire_property$nearest_house <- nearest_house

df_fire_property$diff_time <- df_fire_property$ARRIVAL_DATE_TIME -
  df_fire_property$INCIDENT_DATE_TIME

df_fire_property$diff_time <- as.numeric(df_fire_property$diff_time)

# remove outliers, for more informative graphs

## check for the outliers
head(sort(df_fire_property$min_dist, decreasing  = T)) # one 101812.165 should be removed

[1] 101812.165   4034.679   3543.712   3543.712   3543.712   3543.712

head(sort(df_fire_property$diff_time, decreasing = T)) # two 5339 and 2613 should be removed

[1] 5339 2613 1191 1107 1089 1064

head(sort(df_fire_property$diff_time, decreasing = F)) # one negative number should be removed

[1] -305   12   15   16   17   17

df_fire_property <- df_fire_property[-which(df_fire_property$min_dist > 101812.165),]
df_fire_property <- df_fire_property[-which(df_fire_property$diff_time > 2600),]
df_fire_property <- df_fire_property[-which(df_fire_property$diff_time < 0),]

# minimize the categories of alarms again
df_fire_property$rescale <- ifelse(df_fire_property$HIGHEST_LEVEL_DESC == '1 - More than initial alarm, less than Signal 7-5' | df_fire_property$HIGHEST_LEVEL_DESC =="0 - Initial alarm",
                                   "1", df_fire_property$HIGHEST_LEVEL_DESC)

# set this to factor
df_fire_property$rescale <- factor(df_fire_property$rescale)
levels(df_fire_property$rescale) <- c("less than 2nd alarm", "2nd alarm", "3rd alarm", 
                                      "4th alarm ", "5th alarm", "Signal 7-5")

b) Map of Response Times

pal2 <- brewer.pal(uniqueN(df_fire_property$property), "Paired")
  
residence <- subset(df_fire_property, df_fire_property$property == "Residence")
openA <-  subset(df_fire_property, df_fire_property$property == "Open 
                 ")
dorm <- subset(df_fire_property, df_fire_property$property == "Dorms, Shelters, Hotels")
public <- subset(df_fire_property, df_fire_property$property == "Public 
                 ")
school <- subset(df_fire_property, df_fire_property$property == "Schools")
medical <- subset(df_fire_property, df_fire_property$property == "Medical")


base_map %>%
  addCircleMarkers(data = residence, 
                   group = 'Residence', 
                   lng = ~lon, lat = ~lat, 
                   color = pal2[1],
                   radius = residence$diff_time/100,
                   stroke = 0, weight = 0, fillOpacity = .7, 
                   popup = ~paste0('Address: ', residence$address, '<br/>', 
                            'Incident Data: ', residence$INCIDENT_DATE_TIME, '<br/>',
                            'Total Incident Duration: ',
                            round(residence$TOTAL_INCIDENT_DURATION/60), ' min')) %>% 
  addCircleMarkers(data = openA, 
                   group = 'Open 
                   ', 
                   lng = ~lon, lat = ~lat, 
                   color = pal2[2],
                   radius = openA$diff_time/100,
                   stroke = 0, weight = 0, fillOpacity = .7, 
                   popup = ~paste0('Address: ', openA$address, '<br/>', 
                            'Incident Data: ', openA$INCIDENT_DATE_TIME, '<br/>',
                            'Total Incident Duration: ',
                            round(openA$TOTAL_INCIDENT_DURATION/60), ' min')) %>% 
    addCircleMarkers(data = dorm, 
                   group = 'Dorms, Shelters, Hotels', 
                   lng = ~lon, lat = ~lat, 
                   color = pal2[3],
                   radius = dorm$diff_time/100,
                   stroke = 0, weight = 0, fillOpacity = .7, 
                   popup = ~paste0('Address: ', dorm$address, '<br/>', 
                            'Incident Data: ', dorm$INCIDENT_DATE_TIME, '<br/>',
                            'Total Incident Duration: ',
                            round(dorm$TOTAL_INCIDENT_DURATION/60), ' min')) %>% 
  addCircleMarkers(data = public, 
                   group = 'Public 
                   ', 
                   lng = ~lon, lat = ~lat, 
                   color = pal2[4],
                   radius = public$diff_time/100,
                   stroke = 0, weight = 0, fillOpacity = .7, 
                   popup = ~paste0('Address: ', public$address, '<br/>', 
                            'Incident Data: ', public$INCIDENT_DATE_TIME, '<br/>',
                            'Total Incident Duration: ',
                            round(public$TOTAL_INCIDENT_DURATION/60), ' min')) %>% 
  addCircleMarkers(data = school, 
                   group = 'Schools', 
                   lng = ~lon, lat = ~lat, 
                   color = pal2[5],
                   radius = school$diff_time/100,
                   stroke = 0, weight = 0, fillOpacity = .7, 
                   popup = ~paste0('Address: ', school$address, '<br/>', 
                            'Incident Data: ', school$INCIDENT_DATE_TIME, '<br/>',
                            'Total Incident Duration: ',
                            round(school$TOTAL_INCIDENT_DURATION/60), ' min')) %>% 
  addCircleMarkers(data = medical, 
                   group = 'Medical', 
                   lng = ~lon, lat = ~lat, 
                   color = pal2[6],
                   radius = medical$diff_time/100,
                   stroke = 0, weight = 0, fillOpacity = .7, 
                   popup = ~paste0('Address: ', medical$address, '<br/>', 
                            'Incident Data: ', medical$INCIDENT_DATE_TIME, '<br/>',
                            'Total Incident Duration: ',
                            round(medical$TOTAL_INCIDENT_DURATION/60), ' min')) %>% 
  addLayersControl(overlayGroups = c('Residence','Open 
                                     ',
                                     'Dorms, Shelters, Hotels', 'Public 
                                     ',
                                     'Medical','Schools'),
                   options = layersControlOptions(collapsed = F), 
                   position = "topleft") 

NA

It is really hard to compare the response time among various types of property only by color and size, so I use the checkbox feature in the layers control to make it easier for readers to compare by checking different kinds of property.

levels(df_fire_property$HIGHEST_LEVEL_DESC)[2] <- "1 - More than initial alarm"

pal2 <- brewer.pal(uniqueN(df_fire_property[!df_fire_property$HIGHEST_LEVEL_DESC %in% NA, ]$HIGHEST_LEVEL_DESC), 
                   'YlOrRd')
propCol2 <- colorFactor(palette = pal2, domain = df_fire_property$HIGHEST_LEVEL_DESC)

base_map %>%
  addCircleMarkers(data = df_fire_property[!df_fire_property$HIGHEST_LEVEL_DESC %in% NA, ], 
                   group = 'Incidents', 
                   lng = ~lon, lat = ~lat, 
                   radius = df_fire_property$diff_time/100,
                   color = ~propCol2(HIGHEST_LEVEL_DESC),
                   stroke = 0, weight = 0, fillOpacity = .7, 
                   popup = ~paste0('Address: ', df_fire_property$address, '<br/>', 
                            'Incident Data: ', df_fire_property$INCIDENT_DATE_TIME, '<br/>',
                            'Total Incident Duration: ',
                            round(df_fire_property$TOTAL_INCIDENT_DURATION/60), ' min<br/>',
                            'PropertyType: ', df_fire_property$property)) %>%
  addLegend(data = df_fire_property[!df_fire_property$HIGHEST_LEVEL_DESC %in% NA, ], 
            group = 'Incidents',
            pal = propCol2, values = ~HIGHEST_LEVEL_DESC, 
            title = 'Level of Alarms', position = 'topleft')

NA

nyc <- readOGR('/Users/greatyifan/Desktop/@Columbia/2020spring/2_DataViz/course_materials/Exercises/07_fire/borough_boundaries.geojson')

OGR data source with driver: GeoJSON 
Source: "/Users/greatyifan/Desktop/@Columbia/2020spring/2_DataViz/course_materials/Exercises/07_fire/borough_boundaries.geojson", layer: "borough_boundaries"
with 5 features
It has 4 fields

df_fire_property$year <- year(df_fire_property$ARRIVAL_DATE_TIME)

df_fire_property$boro_name <- gsub('\\d{1}\\s-\\s', '', df_fire_property$BOROUGH_DESC)
df_fire_property$year <- paste0("n_", df_fire_property$year) #avoiding valuable names is number

boro_year_count <- df_fire_property %>% 
  group_by(year, boro_name) %>% 
  count() %>% 
  drop_na() %>% 
  spread(key = 'year', value = 'n')

nyc@data <- nyc@data %>% 
  left_join(boro_year_count, on = "boro_name")

# get centroid for each borough
centers <- as.data.frame(gCentroid(nyc, byid = T))

pal_year <- colorBin("YlOrRd", bins = seq(from = 0, to = 900, by = 100))

map_2013 <- base_map %>% 
  addPolygons(data = nyc, stroke = FALSE, smoothFactor = 0.5,
              weight=1, color='#333333', opacity=1,
              fillColor = ~pal_year(n_2013),
              fillOpacity = .8) %>% 
  addLegend(data = nyc@data, pal = pal_year, position = "topleft", 
            values = ~n_2013, title = "<font size='5'>2013</font>") %>% 
  addLabelOnlyMarkers(lng = centers[,1], lat = centers[,2], 
                      label = paste0(nyc@data$boro_name, "\n", nyc@data$n_2013),
                      labelOptions = labelOptions(noHide = T, textOnly = T, 
                                                  direction = 'center'))

map_2014 <- base_map %>% 
  addPolygons(data = nyc, stroke = FALSE, smoothFactor = 0.5,
              weight=1, color='#333333', opacity=1,
              fillColor = ~pal_year(n_2014),
              fillOpacity = .8) %>% 
  addLegend(data = nyc@data, pal = pal_year, position = "topleft", 
            values = ~n_2014, title = "<font size='5'>2014</font>") %>% 
  addLabelOnlyMarkers(lng = centers[,1], lat = centers[,2], 
                      label = paste0(nyc@data$boro_name, "\n", nyc@data$n_2014),
                      labelOptions = labelOptions(noHide = T, textOnly = T, 
                                                  direction = 'center'))

map_2015 <- base_map %>% 
  addPolygons(data = nyc, stroke = FALSE, smoothFactor = 0.5,
              weight=1, color='#333333', opacity=1,
              fillColor = ~pal_year(n_2015),
              fillOpacity = .8) %>% 
  addLegend(data = nyc@data, pal = pal_year, position = "topleft", 
            values = ~n_2015, title = "<font size='5'>2015</font>") %>% 
  addLabelOnlyMarkers(lng = centers[,1], lat = centers[,2], 
                      label = paste0(nyc@data$boro_name, "\n", nyc@data$n_2015),
                      labelOptions = labelOptions(noHide = T, textOnly = T, 
                                                  direction = 'center'))

map_2016 <- base_map %>% 
  addPolygons(data = nyc, stroke = FALSE, smoothFactor = 0.5,
              weight=1, color='#333333', opacity=1,
              fillColor = ~pal_year(n_2016),
              fillOpacity = .8) %>% 
  addLegend(data = nyc@data, pal = pal_year, position = "topleft", 
            values = ~n_2016, title = "<font size='5'>2016</font>") %>% 
  addLabelOnlyMarkers(lng = centers[,1], lat = centers[,2], 
                      label = paste0(nyc@data$boro_name, "\n", nyc@data$n_2016),
                      labelOptions = labelOptions(noHide = T, textOnly = T, 
                                                  direction = 'center'))

map_2017 <- base_map %>% 
  addPolygons(data = nyc, stroke = FALSE, smoothFactor = 0.5,
              weight=1, color='#333333', opacity=1,
              fillColor = ~pal_year(n_2017),
              fillOpacity = .8) %>% 
  addLegend(data = nyc@data, pal = pal_year, position = "topleft", 
            values = ~n_2017, title = "<font size='5'>2017</font>") %>% 
  addLabelOnlyMarkers(lng = centers[,1], lat = centers[,2], 
                      label = paste0(nyc@data$boro_name, "\n", nyc@data$n_2017),
                      labelOptions = labelOptions(noHide = T, textOnly = T, 
                                                  direction = 'center'))

map_2018 <- base_map %>% 
  addPolygons(data = nyc, stroke = FALSE, smoothFactor = 0.5,
              weight=1, color='#333333', opacity=1,
              fillColor = ~pal_year(n_2018),
              fillOpacity = .8) %>% 
  addLegend(data = nyc@data, pal = pal_year, position = "topleft", 
            values = ~n_2018, title = "<font size='5'>2018</font>") %>% 
  addLabelOnlyMarkers(lng = centers[,1], lat = centers[,2], 
                      label = paste0(nyc@data$boro_name, "\n", nyc@data$n_2018),
                      labelOptions = labelOptions(noHide = T, textOnly = T, 
                                                  direction = 'center'))

leaflet_grid <- 
  tagList(
    tags$table(width = "100%",
               tags$tr(
                 tags$td(map_2013),
                 tags$td(map_2014)
               ),
               tags$tr(
                 tags$td(map_2015),
                 tags$td(map_2016)
               ),
               tags$tr(
                 tags$td(map_2017),
                 tags$td(map_2018)
               )
    )
  )

browsable(leaflet_grid)

Bronx 636 Staten Island 177 Brooklyn 831 Queens 661 Manhattan 545 +− 2013 0 – 100 100 – 200 200 – 300 300 – 400 400 – 500 500 – 600 600 – 700 700 – 800 800 – 900 Leaflet | © Chengwei Wang, © OpenStreetMap contributors © CARTO	Bronx 561 Staten Island 156 Brooklyn 821 Queens 650 Manhattan 561 +− 2014 0 – 100 100 – 200 200 – 300 300 – 400 400 – 500 500 – 600 600 – 700 700 – 800 800 – 900 Leaflet | © Chengwei Wang, © OpenStreetMap contributors © CARTO
Bronx 536 Staten Island 162 Brooklyn 814 Queens 664 Manhattan 538 +− 2015 0 – 100 100 – 200 200 – 300 300 – 400 400 – 500 500 – 600 600 – 700 700 – 800 800 – 900 Leaflet | © Chengwei Wang, © OpenStreetMap contributors © CARTO	Bronx 466 Staten Island 147 Brooklyn 734 Queens 627 Manhattan 479 +− 2016 0 – 100 100 – 200 200 – 300 300 – 400 400 – 500 500 – 600 600 – 700 700 – 800 800 – 900 Leaflet | © Chengwei Wang, © OpenStreetMap contributors © CARTO
Bronx 474 Staten Island 119 Brooklyn 727 Queens 551 Manhattan 426 +− 2017 0 – 100 100 – 200 200 – 300 300 – 400 400 – 500 500 – 600 600 – 700 700 – 800 800 – 900 Leaflet | © Chengwei Wang, © OpenStreetMap contributors © CARTO	Bronx 255 Staten Island 61 Brooklyn 323 Queens 254 Manhattan 219 +− 2018 0 – 100 100 – 200 200 – 300 300 – 400 400 – 500 500 – 600 600 – 700 700 – 800 800 – 900 Leaflet | © Chengwei Wang, © OpenStreetMap contributors © CARTO